Summary: New experiences are encoded by linking them to pre-existing activity patterns in the brain. Memory consolidation occurs when those patterns become connected across brain regions through brief, synchronized bursts of neural activity.
Source: Osaka Metropolitan University
Neuronal ensembles that represent a memory appear to be present in the brain before the experience is even acquired, posing the counterintuitive idea that the brain partially “knows” what it will learn.
A new study investigating how distributed pieces of information are integrated into a single memory provides a plausible explanation for this paradox.
In a fear conditioning paradigm performed in freely behaving rats, researchers recorded large populations of neurons and discovered that ensemble patterns participating in memory formation were already configured in the amygdala and prefrontal cortex before learning, whereas ensembles in the hippocampus were shaped by the experience itself.
“This means that an ensemble’s eventual role or meaning is determined later, once it becomes linked to other ensembles through experience,” says Research Associate Hiroyuki Miyawaki, the study’s lead author.
Working with Professor Kenji Mizuseki from the Department of Physiology, Osaka City University Graduate School of Medicine, the two researchers addressed two seemingly conflicting views about how memories are represented in the brain.
“On one hand, we know that the same neuronal ensemble often becomes active during both memory acquisition and later recall,” Professor Mizuseki explains. “On the other hand, prevailing models propose that new information is first stored as short-term traces in areas like the hippocampus and then gradually transferred to cortical regions during sleep to form long-term memories.”
The team realized these differing observations arose from examining memory at different anatomical scales. To reconcile them, they needed to record activity from many neurons across multiple regions at once and analyze both local and interregional ensemble dynamics — a technically demanding task.
To achieve this, the researchers used large-scale electrophysiological recordings simultaneous in the amygdala, hippocampus, and prefrontal cortex—regions known to be critical for fear memory—in freely moving rats. They then applied mathematical analyses to identify ensemble firing patterns in each region during the fear conditioning task and quantified the activation strength of each ensemble.
Finally, they examined the temporal relationships among ensembles across the three brain regions to determine how coordination evolved during learning, subsequent sleep, and later recall.
Their analysis produced three main findings that clarify how distributed memory elements are organized and bound together.
First, although ensemble synchrony linking amygdala–prefrontal and hippocampus–prefrontal pairs was observed around the time of learning, the timing of their development differed. Amygdala–prefrontal coordination was already evident during acquisition, while hippocampus–prefrontal coordination was weak at that stage and became stronger later, particularly during post-learning sleep. In contrast, both types of cross-regional synchrony were clearly present during later memory recall.
“When we compared the proportion of ensemble pairs showing significant coactivity during acquisition and recall,” Miyawaki notes, “amygdala–prefrontal ensemble interactions tended to decrease over time, whereas hippocampus–prefrontal interactions increased.”
These results suggest the amygdala–prefrontal network can form rapidly during an experience, whereas the hippocampus–prefrontal network develops more gradually and depends on the subsequent processing of that experience.
Second, the team observed that interactions among ensembles across regions were particularly strong during transient, high-frequency bursts of activity. Examples include hippocampal ripple oscillations, high-frequency events in the amygdala, and ripple-like oscillations in the prefrontal cortex. These brief network bursts were prominent during post-learning sleep and reappeared during memory retrieval.
This finding is consistent with previous work linking transient oscillatory bursts to memory consolidation during sleep and to memory-related processing during wakefulness, and it indicates that coordinated, cross-regional ensemble activity tied to these bursts plays an essential role in forming accessible memories.
Third, the researchers found that local neuronal ensembles in the amygdala and prefrontal cortex were present before the conditioning experience, whereas hippocampal ensembles emerged in an experience-dependent way. Importantly, the synchronized coactivation of ensembles spanning regions was not observed prior to learning.
Taken together, these observations indicate that information about a new experience is rapidly incorporated by linking it to pre-existing activity patterns in the amygdala and prefrontal cortex. The wider cross-regional network that integrates these distributed elements, however, forms more slowly and relies on activity patterns that develop with experience, especially during periods of sleep-associated replay and transient oscillatory bursts.
“Our results suggest that new information is retained by being bound to preconfigured activity patterns, and that a memory is established when those patterns are connected across brain regions through brief bouts of synchronized activity,” concludes Professor Mizuseki.
Because the fear conditioning paradigm models aspects of human post-traumatic stress disorder (PTSD), the authors hope these findings will inform future approaches to treating memory-related disorders. They also plan to extend this line of research beyond fear learning to uncover general principles of memory operation and to clarify how memory function deteriorates with aging and disease.
About this memory research news
Author: Press Office
Source: Osaka Metropolitan University
Contact: Press Office – Osaka Metropolitan University
Image: The image is in the public domain
Original Research: Open access.
“De novo inter-regional coactivations of preconfigured local ensembles support memory” by Hiroyuki Miyawaki and Kenji Mizuseki. Nature Communications
Abstract
De novo inter-regional coactivations of preconfigured local ensembles support memory
Neuronal ensembles in the amygdala, ventral hippocampus, and prefrontal cortex contribute to fear memory, yet how interactions among these region-specific ensembles support memory has remained unclear. Using multi-regional, large-scale electrophysiological recordings in fear-conditioned rats, the study shows that locally active ensembles during acquisition are later coactivated across regions during post-training sleep, a process that depends on brief episodes of fast network oscillations.
During later memory retrieval, the same inter-regional coactivations reappeared alongside fast oscillatory events. Ensembles that participated in these coactivations were preconfigured in the amygdala and prefrontal cortex before learning, while hippocampal ensembles developed through experience. These results indicate that elements of a new memory are immediately encoded within multiple brain regions in a preconfigured manner, whereas hippocampal ensembles and the inter-regional network that integrates distributed information mature in an experience-dependent fashion, in line with the hippocampal memory index hypothesis.